The invisible barrier
The “New Space” gold rush of the 2020s was built on a single, silent assumption: that Low Earth Orbit (LEO) would remain a benign environment. If companies could launch cheap enough and fast enough, the physics of the vacuum would take care of the rest. That assumption took two hits in one week of December 2025, and the two hits had nothing to do with each other.
On December 17, 2025, Starlink satellite 35956 lost contact with the ground at 418 kilometers. SpaceX’s own account: “The anomaly led to venting of the propulsion tank, a rapid decay in semi-major axis by about 4 km, and the release of a small number of trackable low relative velocity objects.” The tracking firm LeoLabs counted it differently. Within days it was following “hundreds” of debris objects from the event, spread across 6,000 kilometers of the orbital track, while the satellite itself stayed intact and tumbling.
Five days later, on December 22, the Associated Press reported that intelligence services from two NATO countries assess that Russia is developing a “zone-effect” weapon meant to flood Starlink’s operating altitude, around 550 kilometers, with “hundreds of thousands of high-density pellets,” disabling many satellites at once rather than targeting one. The assessment did not say when Russia could field such a system or whether it had been tested.
The Starlink failure was an internal propulsion problem, not an attack. But read together, the two stories describe the same exposure from opposite directions: a fleet of thousands of cheap satellites is vulnerable to things too small to see, whether they come from a leaking tank or a hostile launch. That exposure is what this article calls the LEO insurance cliff. It is the point where the ability to launch outruns the ability to underwrite, and the underwriters were already wobbling: a single geostationary loss in late December pushed the 2025 space-insurance loss ratio “from circa 15% to circa 75%,” while the cheapest new cover for a debris strike starts around $30,000 a year and exists only because a sensor can now prove what hit you.
Whipple shields versus the pellet
Most satellites protect themselves with some version of a Whipple shield, an idea Fred Whipple proposed in 1947. A thin outer bumper sits a short distance from the main hull. When debris hits the bumper, “the bumper is not expected to stop the incoming particle or even remove much of its energy, but to break up and disperse it, dividing the original particle energy among many fragments that fan out between bumper and wall.”
That works against the debris the shield was sized for, which for typical spacecraft shielding means particles up to about 1 centimeter. The problem is everything between that and what the catalog can see. The European Space Agency’s (ESA) 2026 Space Environment Report, using its MASTER reference population as of February 1, 2026, estimates 68,450 objects larger than 10 centimeters in orbit, including about 11,300 active payloads, plus 1.5 million debris objects between 1 and 10 centimeters and 230 million between 1 millimeter and 1 centimeter. The tracked population is the 68,450. The other 231 million or so are invisible to the catalog and lethal anyway.
A 1-millimeter pellet is nothing until you multiply by orbital velocity. Kinetic energy is:
For a 1-gram pellet at 7.5 kilometers per second:
A .50 BMG rifle round leaves the muzzle with between about 14,000 and 20,000 joules depending on the load. A gram of metal in orbit carries more than that, and a weapon that seeds “hundreds of thousands” of them does not need to aim. The Whipple defense becomes a matter of statistics, not engineering. Nobody is targeted; the orbit is denied.
Every impact then feeds the next. ESA’s report puts it bluntly: combined with the constellations being launched, “this has the Kessler syndrome’s runaway effect becoming visible much earlier in time,” and the agency has shortened its prediction horizon from 200 years to 100 because the projected numbers grew too large to simulate.
Would a pellet weapon even work?
The experts AP called were split, and the split is worth reading closely. Victoria Samson of the Secure World Foundation, a nonprofit that tracks counterspace programs: “I don’t buy it. Like, I really don’t,” though she added, “I wouldn’t put it past some scientists to build out something like this.” Brigadier General Christopher Horner of the Canadian military’s space division said “it’s not implausible.” Clayton Swope of the Center for Strategic and International Studies: “If satellites start winking out with damage, I guess you could put two and two together.” The Kremlin did not respond to AP’s request for comment.
The skeptics’ case is the physics above, run in reverse. A pellet cloud at 550 kilometers does not check passports. Russia’s and China’s satellites fly through the same shells, and so would any future Russian constellation. That is the same reason no country has used a destructive weapon on another’s satellite in a real conflict. The Secure World Foundation’s 2026 assessment of 13 countries’ programs found that “only non-destructive capabilities are actively being used against satellites in current military operations.”
The clearest signal since January came on September 14, 2026, when Air Force Secretary Troy Meink told the Air & Space Forces Association conference: “This is why the United States now has on-orbit space control weapons capable of defending the joint force against hostile adversary action.” He would not say what they are. Samson’s read: “I’m going to guess what he’s talking about is not a kinetic space-control weapon due to multiple comments government officials have said about avoiding debris causing counter space.” Her guess: “space-based jammers, electronic warfare.” The full account of what was said and withheld is worth reading alongside this piece. The point for orbit is that the weapons the United States chose to acknowledge are, by the best outside guess, the reversible kind. The pellet concept runs the other way, which is exactly why the people who track this doubt it.
The underwriting crisis
Insurance was always the hidden cost of space, and it was built for the old model: a handful of very expensive satellites in geostationary orbit, each insured individually. That market just took a beating. On December 26, 2025, underwriters received a notice of potential loss for SpainSat NG-2, insured for more than $400 million, and by the time the broker Gallagher published its first-quarter 2026 update the total loss was apparently confirmed, moving the 2025 space-insurance loss ratio “from circa 15% to circa 75%.” Capacity still rose by about $100 million year on year, to roughly $300 million of competitive capacity, but Gallagher expects 2026 premium to “significantly contract from 2025,” partly because of “a continuing lack of buyer appetite for insuring satellite constellations.”
That last clause is the cliff. Constellation operators have mostly not bought conventional cover. A fleet of thousands of identical, replaceable satellites is cheaper to self-insure than to insure, and that logic holds when the risk is random collision. It fails when the risk is deliberate denial of an entire altitude: every satellite in the shell is exposed at once, and nothing is replaceable fast enough.
The Arkisys-Odin pivot
On December 22, 2025, the same day as the AP story, Arkisys and Odin Space announced what they called the first collision-specific satellite insurance, built around Odin’s nano sensors, which record sub-centimeter impacts on a spacecraft’s surface and act as a “black box” for the strike. Odin’s chief executive, James New: “Our partnership with Arkisys marks a pivotal shift in how the industry tackles the rising threat of lethal, non-trackable debris.”
Why does a sensor matter to an insurer? Because without data there is no claim. If a satellite dies in LEO with no witness, the underwriter can call it a workmanship failure or a solar flare. Verified telemetry of a debris strike turns an unprovable loss into a payable one. As of September 2026 Odin’s site says its coverage “starts around $30,000 a year, a fraction of the cost of traditional, all-peril, policies,” and is “underwritten by Lloyd’s of London.” Whether that price survives a real pellet cloud is the open question; it is priced for accidents, and no one has yet priced a deliberate one.
From ASAT to denied orbits
The crude version of orbital denial already happened. On November 15, 2021, Russia’s A-235 Nudol interceptor destroyed the defunct Kosmos 1408 satellite, leaving “about 1,500 pieces of debris that can be tracked by ground-based radar” and “hundreds of thousands more that are more difficult to track,” by the US State Department’s count, and sending the seven people on the International Space Station into their docked spacecraft in case the station were hit. A pellet weapon would be that event without the explosion: no single flash to attribute, just a shell that quietly stops being safe.
The margin that used to absorb such events is gone. ESA’s 2026 report says its latest reference population “reveals that the density of active payloads is approaching that of space debris in heavily populated altitude bands for the first time,” after a 2025 in which more than 300 launches put over 4,000 new payloads into orbit. The December Starlink failure showed what that density means in practice: one satellite’s bad day became hundreds of tracked objects strung across thousands of kilometers.
The fragmented sky
The “roaming” dream of a single, seamless global satellite network is splitting into tiers.
- The hardened tier: Military and high-value commercial assets add mass for shielding. More shielding means heavier satellites, higher launch cost per unit of bandwidth.
- The uninsurable tier: Cheap CubeSats and academic missions get pushed to lower, shorter-lived orbits where atmospheric drag clears debris faster and mission life drops from years to months.
- The sovereignty tier: Nations that treat orbit as national infrastructure (see The Silicon Swadeshi) will fund protected orbits and active debris removal (see The Orbital Janitor).
The second-order effect is a broadband tax. If constellation operators are eventually forced to buy sensor-verified cover priced per satellite, that cost has one place to go: the monthly bill of the rural subscriber refreshing a browser in a dead zone.
The “smart friend” summary
If an argument breaks out at a bar, remember three numbers: 7.5, 10, and 230 million.
- 7.5: kilometers per second, the speed that turns a gram of metal into something with more energy than a .50-caliber round.
- 10: centimeters, the size below which the catalog stops counting; shields are typically built for hits of about 1 centimeter, and the gap between those two numbers holds 1.5 million objects.
- 230 million: ESA’s estimate of debris objects between 1 millimeter and 1 centimeter already up there, before anyone launches a single pellet.
Low Earth Orbit is no longer just a technical frontier; it is a financial one, and the underwriters are the ones standing at the edge.
Correction and update, September 15, 2026: an earlier version of this article dated the Starlink anomaly to December 18; SpaceX’s account puts it on December 17. It also described the NATO-country intelligence as a leaked briefing within 48 hours of the anomaly; AP reported it on December 22 and the two events are unrelated. A 400 percent premium forecast that had no source and an incorrect link for the intelligence reporting were removed, and the debris, insurance-market and counterspace figures were refreshed against ESA’s 2026 report, Gallagher’s Q1 2026 update and the Secure World Foundation’s 2026 assessment.
Sources (13)
- x.com Starlink on X: December 17 anomaly on satellite 35956
- theregister.com Starlink satellite fails, polluting orbit with debris and falling toward Earth
- pbs.org PBS NewsHour (AP): Intelligence agencies suspect Russia is developing anti-satellite weapon to target Starlink service
- interestingengineering.com World's First Space Debris Collision Insurance Launched
- odin.space Real-Time Debris Detection & Insurance
- specialty.ajg.com Gallagher: Space Insurance market update Q1 2026
- sdo.esoc.esa.int ESA Space Debris Office: Annual Space Environment Report 2026 (PDF)
- cdn.prod.website-files.com Secure World Foundation: Global Counterspace Capabilities 2026
- defenseone.com US military has 'on-orbit' space weapons, Air Force Secretary reveals
- twz.com The War Zone: U.S. Admits It Has Weapons In Orbit
- en.wikipedia.org Wikipedia: Kosmos 1408
- en.wikipedia.org Wikipedia: Whipple shield
- en.wikipedia.org Wikipedia: .50 BMG
🦋 Discussion on Bluesky
Discuss on Bluesky